On Friday, the 24th, Meta announced a significant improvement to its artificial intelligence chatbot. This update expands the tool's capabilities beyond existing functions, such as generating responses, creating images, and drafting documents, by incorporating new features focused on personal organization, detailed research, and task planning.
The new feature allows the assistant to access the user's calendar data to generate daily summaries, assist in selecting dates for appointments, and support overall activity organization. Meta claims that these additions are part of a larger strategy to transform its AI into a true personal assistant.
The initial launch is available in specific markets, accessible through both the Meta AI app and the service's web version. The company communicated that expansion to other regions and platforms, including WhatsApp, will be implemented in the coming weeks.
Meta presented this update as an evolution of the chatbot, which now includes tools to monitor repetitive tasks and suggest actions based on context provided by the user. It is possible to configure the system once so that it executes certain functions continuously, without the need for constant commands.
Among the practical examples cited by the company are the creation of weekly meal schedules, alerts about the need to restock items, and the drafting of daily agenda summaries. Furthermore, the assistant can help with practical searches, such as locating used furniture on Facebook Marketplace while respecting a predefined budget for renovations.
The tool has also received dedicated features for events and searches. According to Meta, the chatbot has the ability to find restaurants suitable for a specific occasion and cross-reference this information with the user's schedule to identify a compatible day. Additionally, the system now produces reports, presentations, and plans using information collected from the internet. Another innovation allows users to redirect the search while the answer is still being generated, following a pattern seen in other AI assistants.
This update is powered by the Muse Spark 1.1 model, which Meta designated as responsible for increasing the chatbot's abilities. The company associated this novelty with the concept of 'personal superintelligence,' a term used by CEO Mark Zuckerberg when discussing the future of artificial intelligence.
This change signals a shift in the company's previous approach. According to information attributed to journalist Alex Heath, Chris Cox, Meta's product director, had instructed employees last year to focus on entertainment and social connection, rather than productivity—an area where competitors like OpenAI, Google, and Anthropic are strong. With these new features, Meta enters a more direct competition with the AI assistants developed by its rivals, seeking to increase the daily use of the tool through integration with organization, research, and planning tasks.
According to sources familiar with the matter, preliminary negotiations are underway regarding a potential deal between Meta and Anthropic concerning the lease of part of Meta's computing infrastructure, which is valued in billions of dollars.
The New York Times initially reported on this possible agreement, citing three informed individuals who estimated the deal's value at up to ten billion dollars over two years. However, a CNN source familiar with the situation noted that any specific figures published are merely speculation.
Both Meta and Anthropic declined to comment on these negotiations.
The possibility of becoming a provider of computing power could open up a significant new revenue stream for Meta, as the company is actively investing in developing data center infrastructure to support its artificial intelligence goals. According to the latest earnings report, Meta plans to spend between $125 and $145 billion on capital expenditures this year, primarily directed towards expanding this infrastructure. This amount could double the company's spending compared to the previous year.
Earlier, in April, Meta announced a 10% workforce reduction, affecting about 8,000 employees, partly to offset costs associated with these investments.
Meta CEO Mark Zuckerberg has previously spoken about the possibility of leasing out part of this infrastructure if the company's own computing needs cannot keep pace with construction rates. At Meta's annual shareholder meeting in May, Zuckerberg stated: 'We get inquiries almost every week from various external companies asking if we have computing power they could purchase from us at a premium price compared to the purchase price. We haven't done that yet because we believe we have uses for this computing. But obviously, if we reach a point where we feel we have built too much, then it is one of the options we have.'
Demand for computing power remains high as both large and small companies rush to implement AI, and leading AI labs work on improving their models. Anthropic has already secured multi-billion dollar agreements for computing power licensing with companies such as Google, SpaceX, Microsoft, and Amazon.
Meanwhile, investors are waiting to see how Meta's investments will affect net profit, especially given the need to keep up with AI offerings from companies like Anthropic and OpenAI. Meta's stock (META) has fallen by more than 8 percent compared to this time last year. Furthermore, last month, Meta released an updated version of its Muse Spark AI model, stating that it can compete with the coding capabilities of models from OpenAI, Anthropic, and others. For the first time, Meta announced plans to offer this service in a paid version, which is another sign that the company is looking for higher returns on its AI developments.
